A Human Lung Carcinoma Cell Line Supports Efficient Measles Virus Growth and Syncytium Formation via a SLAM- and CD46-Independent Mechanism

Abstract

Measles virus (MV) propagates mainly in lymphoid organs throughout the body and produces syncytia by using signaling lymphocyte activation molecule (SLAM) as a receptor. MV also spreads in SLAM-negative epithelial tissues by unknown mechanisms. Ubiquitously expressed CD46 functions as another receptor for vaccine strains of MV but not for wild-type strains. We here show that MV grows and produces syncytia efficiently in a human lung adenocarcinoma cell line via a SLAM- and CD46-independent mechanism using a novel receptor-binding site on the hemagglutinin protein. This infection model could advance our understanding of MV infection of SLAM-negative epithelial cells and tissues.

Measles is an acute, contagious disease characterized by high fever, cough, and a maculopapular rash (8). The etiologic agent is Measles virus (MV), which belongs to the genus Morbillivirus in the family Paramyxoviridae. MV initiates its infectious cycle by attaching the hemagglutinin (H) protein on the virus envelope to a cellular receptor on a target cell. Attachment of the H protein to a receptor triggers membrane fusion between the virus envelope and the plasma membrane of the target cell mediated by the fusion (F) protein (14). The signaling lymphocyte activation molecule (SLAM) (also known as CD150) and CD46 have been identified as receptors for MV (5, 6, 10, 22, 47). SLAM is a common receptor for all strains of MV, whereas CD46 functions as a receptor for only vaccine strains and some laboratory strains of MV (52). Ono et al. showed that MV strains circulating in patients (wild-type [wt] MV strains) use SLAM but not CD46 (26). Formation of syncytia is characteristic of MV-infected cells (8). SLAM is also required for this process by wt MV (52).

Pathological examination of patients and monkeys infected with MV has indicated that lymphoid organs are major targets of MV (3, 23, 28, 49, 51), and the distribution of SLAM is well correlated with sites of MV spread in vivo (52). Pathological data also show that MV antigens and syncytia are detected in epithelial tissues in various organs, such as the skin, esophagus, oral mucosa, trachea, intestines, pharynx, and urinary bladder (4, 12, 15, 16, 20, 23-25, 28, 34). Therefore, epithelial tissues are likely targets of MV, as are lymphoid organs in vivo. Previous studies using a panel of cell lines showed that only SLAM-positive cells support efficient wt MV infection and syncytium formation (43, 46). Although studies have shown a low level of SLAM-independent infection by wt MV in various cell lines (the efficiency was 100 to 1,000 times lower than that using SLAM), this type of infection usually does not produce syncytia in infected cells (9). Exceptions are primary cultures of human small airway epithelial cells and endothelial cells (1, 43). wt MV induced syncytia in these cells via a SLAM-independent mechanism (1, 43). However, experiments using primary cultures of cells are difficult to reproduce because of significant variations in their preparation. In this paper, we identified and characterized the human lung adenocarcinoma cell line NCI-H358 (2), which supports wt MV entry, replication, and syncytium formation via a novel mechanism.

NCI-H358 is derived from the lung tumor tissue (adenocarcinoma) obtained in 1981 from a male patient prior to initiation of chemotherapy (2). NCI-H358 expresses a major lung surfactant-associated protein, SP-A, and has cytoplasmic granules characteristic of Clara cells (7). The replication kinetics of IC323-EGFP, a recombinant wt MV expressing enhanced green fluorescent protein (9, 42), were analyzed with NCI-H358, Vero/hSLAM (26), B95a (13), and Vero cells. In Vero/hSLAM and B95a cells, the virus grew very efficiently, whereas it hardly grew in Vero cells (Fig. 1A), as reported previously (9, 13, 19, 31, 35, 44). In NCI-H358 cells, IC323-EGFP did replicate and produce syncytia, though at a lower rate than in Vero/hSLAM or B95a cells (Fig. 1A and B). After growing in NCI-H358 cells, IC323-EGFP retained its original cell tropism of wt MV (data not shown). The cell infectious unit (CIU) of a stock of IC323-EGFP was determined for these cell lines according to a protocol described previously (40). It was found to be 10⁶ CIU/ml for Vero/hSLAM cells and 1.5 × 10⁵ CIU/ml for B95a cells (Fig. 1C). Both Vero/hSLAM and B95a cells express high levels of SLAM (26, 47). The CIU for Vero cells was much lower than that for Vero/hSLAM cells. These results are consistent with those of previous studies (9, 26, 36, 44). Under these conditions, the same stock of IC323-EGFP exhibited 8 × 10⁴ CIU/ml in NCI-H358 cells (Fig. 1C), indicating that NCI-H358 cells are susceptible and permissive to wt MV at a level similar to that of B95a cells. These observations were confirmed using six clinical isolates of wt MV belonging to genotypes D3 and D5 (37, 41). All six strains grew and produced syncytia in NCI-H358 cells, as they do in SLAM-positive cell lines (data not shown).

FIG. 1.

wt MV infection of NCI-H358. (A) Replication kinetics. NIC-H358, Vero/hSLAM, B95a, and Vero cells were infected with IC323-EGFP at a multiplicity of infection of 0.01. At various intervals, the CIU was determined for Vero/hSLAM cells. p.i., postinfection. (B) NCI-H358 cells were infected with IC323-EGFP at an MOI of 0.05 and observed at 1, 2, and 3 days p.i. (d.p.i.) by fluorescence (green fluorescent protein [GFP]) and phase-contrast (Phase) microscopy. (C) Infectivity for different cell lines. The infectious titer (CIU) of a stock of IC323-EGFP was determined for NIC-H358, Vero/hSLAM, B95a, and Vero cells.

Monoclonal antibody (MAb) against SLAM (IPO-3) (33) (Kamiya Biomedical, Seattle, WA) inhibited IC323-EGFP infection of Vero/hSLAM cells, as shown previously (31, 47), whereas the same amount of IPO-3 did not inhibit IC323-EGFP infection of NCI-H358 cells (Fig. 2A). These observations were confirmed using the Renilla luciferase-expressing recombinant wt MV, IC323-Luci, which was generated using reverse-genetics protocols (21, 38, 39). The Renilla luciferase gene was inserted into the 3′ first locus of the virus genome, as was done for IC323-EGFP (9). IPO-3 completely blocked IC323-Luci infection of Vero/hSLAM cells but not that of NCI-H358 cells (Fig. 2B). Indeed, microarray analysis and reverse transcription-PCR experiments showed a negligible level of SLAM mRNA in NCI-H358 cells, and flow cytometry showed an undetectable level of SLAM expression on NCI-H358 cells (data not shown). All data indicate that wt MV infection of NCI-H358 cells is SLAM independent.

FIG. 2.

SLAM-independent infection and syncytium formation. (A) Monolayers of Vero/hSLAM and NCI-H358 cells were infected with IC323-EGFP. Some MV-infected monolayers were incubated in medium containing a MAb directed against SLAM (IPO-3). At 36 h postinfection, the monolayers were observed under a fluorescence microscope. (B) Monolayers of Vero/hSLAM and NCI-H358 cells were infected with IC323-Luci. Some MV-infected monolayers were incubated in medium containing IPO-3. At 36 h postinfection, the Renilla luciferase activity in the cells was analyzed by using a Renilla luciferase assay system (Promega, Madison, WI) with a Mithras LB940 plate reader (Berthold Technologies, Pforzheim, Germany). The activity in cells cultured in the absence of IPO-3 was set to 100%. Bars indicate the means ± standard deviations for triplicate samples. (C) Monolayers of NCI-H358 or B95a cells were transfected with an H protein-encoding plasmid (pCA7-ICH, pCA7-ICH-R533A, or pCA7-ICH-Y553A) and an F protein-encoding plasmid (pCXN2-EdF) using the Lipofectamine 2000 reagent (Invitrogen Life Technologies, Carlsbad, CA). Some monolayers were transfected with only pCXN2-EdF [(−)]. At 2 days posttransfection, the cells were stained with a Giemsa solution and observed under a light microscope.

R533A and Y553A substitutions of the H protein are known to disrupt the interaction of the H protein with SLAM (48). We analyzed the fusion support activity of the H protein with these substitutions in NCI-H358 cells. When the wt MV H protein (ICH) was expressed together with the F protein using the eukaryotic expression vector pCA7 (40), many syncytia developed in NCI-H358 cells as well as in B95a cells (Fig. 2C, ICH). No syncytium was detected when only the F protein was expressed [(Fig. 2C, (-)]. The mutant ICH protein with R533A or Y553A failed to support cell-cell fusion in B95a cells, as reported previously (48). In contrast, mutant ICH proteins with R533A or Y553A were able to support cell-cell fusion in NCI-H358 cells (Fig. 2C, R533A and Y553A). These results indicate that the wt MV H protein can support cell-cell fusion in NCI-H358 cells via a SLAM-independent mechanism.

IC/EdH-EGFP possessing the H protein of the Edmonston vaccine strain (EdH) has been shown to grow efficiently in Vero cells using CD46 as a receptor (9, 31, 36). A MAb against CD46, M75 (a gift from T. Seya) (32), blocked IC/EdH-EGFP infection of Vero cells, whereas it showed no effect on the infection of NCI-H358 cells with IC/EdH-EGFP (Fig. 3A). These observations were confirmed with the recombinant MV (IC/EdH-Luci), which possesses the EdH protein and expresses the Renilla luciferase. M75 completely blocked IC/EdH-Luci infection of Vero cells but showed no effect on the infection of NCI-H358 cells with IC/EdH-Luci (Fig. 3B). These findings indicate that IC/EdH-EGFP and IC/EdH-Luci infect NCI-H358 cells via a CD46-independent mechanism. Analyses by flow cytometry using MAbs against CD46 indicated that NCI-H358 expresses CD46 on the cell surface (data not shown). Nonetheless, blockage of CD46 did not affect these virus infections. Therefore, CD46-independent infection of NCI-H358 with IC/EdH-EGFP and IC/EdH-Luci must be as efficient as that using CD46.

FIG. 3.

CD46-independent infection. (A) Monolayers of Vero and NCI-H358 cells were infected with IC/EdH-EGFP. Some monolayers were incubated in medium containing a MAb directed against CD46 (M75) and the others in the absence of M75 (No Ab). At 36 h postinfection, the monolayers were observed under a fluorescence microscope. (B) Monolayers of Vero and NCI-H358 cells were infected with IC/EdH-Luci. Some monolayers were incubated in medium containing a MAb directed against CD46 (M75) and the others in the absence of M75 [(−)]. At 36 h postinfection, the Renilla luciferase activity in the cells was analyzed by using a Renilla luciferase assay system with a Mithras LB940 plate reader. The activity in cells cultured in the absence of M75 was set to 100%. Bars indicate the means ± standard deviations for triplicate samples.

The neutralizing activities of four MAbs against the MV H protein (E81, E103, E396, and E500) were determined with NCI-H358, B95a, and Vero cells. These MAbs were generated as described previously (29, 30). Infection of both B95a and NCI-H358 cells with IC323-EGFP or IC323-Luci was neutralized by E81, E103, and E396 (Fig. 4A and B). E500 showed no effect when B95a cells were infected with IC323-EGFP or IC323-Luci (Fig. 4A and B), but it partially neutralized IC323-EGFP and IC323-Luci infection of NCI-H358 cells (Fig. 4A and B). These results indicate that the receptor-binding site on the H protein used to infect NCI-H358 cells is different from that for SLAM. E81 and E103 also efficiently neutralized IC/EdH-EGFP and IC/EdH-Luci virus infection of both Vero and NCI-H358 cells (Fig. 4A and C). E500 partially neutralized IC/EdH-EGFP and IC/EdH-Luci infection of Vero and NCI-H358 cells (Fig. 4A and C). The effects of E396 differed in Vero and NCI-H358 cells. It completely neutralized IC/EdH-EGFP and IC/EdH-Luci infection of NCI-H358 cells but only partially neutralized that of Vero cells. (Fig. 4A and C). These findings suggest that the receptor-binding site on the H protein used to infect NCI-H358 cells is different from that for CD46.

FIG. 4.

Neutralizing activities of MAbs directed against the MV H protein in NCI-H358, B95a, and Vero cells. (A) Monolayers of NCI-H358, B95a, and Vero cells were infected with IC323-EGFP or IC/EdH-EGFP and cultured in medium containing MAbs directed against the MV H protein (E81, E103, E396, or E500) or medium without Abs [(−)]. At 2 days postinfection, the monolayers were observed under a fluorescence microscope. (B) Monolayers of NCI-H358 or B95a cells were infected with IC323-Luci and cultured in medium containing E81, E103, E396, or E500 or medium without Abs [(−)]. At 2 days postinfection, the Renilla luciferase activity in the cells was analyzed by using a Renilla luciferase assay system with a Mithras LB940 plate reader. The activity in cells cultured in the absence of MAb was set to 100%. Bars indicate the means ± standard deviations for triplicate samples. (C) Monolayers of NCI-H358 or Vero cells were infected with IC/EdH-Luci and cultured in medium containing E81, E103, E396, or E500 or medium without antibody [(−)]. At 2 days postinfection, the Renilla luciferase activity in the cells was analyzed. The activity in cells cultured in the absence of MAb was set to 100%. Bars indicate the means ± standard deviations for triplicate samples.

Our data indicate that NCI-H358 is susceptible to wt MV. In addition, all data obtained using various methods indicate that the wt MV infection and syncytium formation observed in NCI-H358 cells are SLAM independent. This is the first report identifying a cell line that supports wt MV infection efficiently via a SLAM-independent pathway. Manchester et al. reported the possible use of CD46 by some clinical isolates of MV (17), though much evidence shows that wt MV strains isolated using SLAM-positive cells generally do not use CD46 as a receptor (26, 52). We also investigated whether wt MV uses CD46 as a receptor to infect NCI-H358 cells. In these cells, neither wt MV infection nor cell-cell fusion was affected by M75. Furthermore, the 485VP/SS substitution, which has been shown to disrupt binding of the H protein to CD46 (48), also did not affect cell-cell fusion in NCI-H358 cells (data not shown). These results confirmed that CD46 is not involved in wt MV infection and cell-cell fusion in NCI-H358 cells. Our data also show that the Edmonston vaccine strain can infect NCI-H358 cells via a SLAM- and CD46-independent pathway. Thus, wt MV has an intrinsic ability to use two receptors efficiently (one is SLAM, and the other is unknown but is present on NCI-H358), whereas vaccine strains of MV can use three molecules as receptors (CD46 and the two used by wt MV). The infection of NCI-H358 with MV could likely be a model of MV infection of epithelial cells. However, it remains to be elucidated whether a receptor molecule on NCI-H358 is the same as that on small airway epithelial cells or primary cultures of endothelial cells.

Previous studies using a series of mutant MV H proteins and MAbs against the MV H protein have indicated that the binding sites for SLAM and CD46 are located in different positions on the H protein (18, 48). Our findings suggest that the binding site for the molecule on NCI-H358 cells is located in a different position from those for SLAM and CD46. However, mutation analyses of the MV H protein suggest that the binding site for a receptor on NCI-H358 cells may overlap with that for CD46, since the presence of tyrosine at position 481, which is critical for the interaction between the H protein and CD46 (11, 45, 50), increased MV replication and syncytium formation in NCI-H358 cells (data not shown). Vaccine strains of MV are believed to have acquired the N481Y substitution in order to use CD46 (27), but our findings suggest another possibility, that MV might have acquired this substitution to efficiently use another receptor on NCI-H358 cells, which might be expressed on other epithelial cells used to grow MV vaccines. Studies to identify the receptor molecule for MV on NCI-H358 cells are in progress in our laboratory and would greatly advance our understanding of the molecular basis of MV spread in cultured cells and in vivo.